fltk/src/Fl_Bitmap.cxx

529 lines
16 KiB
C++
Raw Normal View History

//
// "$Id$"
//
// Bitmap drawing routines for the Fast Light Tool Kit (FLTK).
//
// Copyright 1998-2008 by Bill Spitzak and others.
//
// This library is free software; you can redistribute it and/or
// modify it under the terms of the GNU Library General Public
// License as published by the Free Software Foundation; either
// version 2 of the License, or (at your option) any later version.
//
// This library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
// Library General Public License for more details.
//
// You should have received a copy of the GNU Library General Public
// License along with this library; if not, write to the Free Software
// Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
// USA.
//
// Please report all bugs and problems on the following page:
//
// http://www.fltk.org/str.php
//
/** \fn Fl_Bitmap::Fl_Bitmap(const char *array, int W, int H)
The constructors create a new bitmap from the specified bitmap data.*/
/** \fn Fl_Bitmap::Fl_Bitmap(const unsigned char *array, int W, int H)
The constructors create a new bitmap from the specified bitmap data.*/
#include <FL/Fl.H>
#include <FL/x.H>
#include <FL/fl_draw.H>
#include <FL/Fl_Widget.H>
#include <FL/Fl_Menu_Item.H>
#include <FL/Fl_Bitmap.H>
#include "flstring.h"
#ifdef __APPLE_QD__ // MacOS bitmask functions
Fl_Bitmask fl_create_bitmask(int w, int h, const uchar *array) {
Rect srcRect;
srcRect.left = 0; srcRect.right = w;
srcRect.top = 0; srcRect.bottom = h;
GrafPtr savePort;
GetPort(&savePort); // remember the current port
Fl_Bitmask gw;
NewGWorld( &gw, 1, &srcRect, 0L, 0L, 0 );
PixMapHandle pm = GetGWorldPixMap( gw );
if ( pm )
{
LockPixels( pm );
if ( *pm )
{
uchar *base = (uchar*)GetPixBaseAddr( pm );
if ( base )
{
PixMapPtr pmp = *pm;
// verify the parameters for direct memory write
if ( pmp->pixelType == 0 || pmp->pixelSize == 1 || pmp->cmpCount == 1 || pmp->cmpSize == 1 )
{
static uchar reverse[16] = /* Bit reversal lookup table */
{ 0x00, 0x88, 0x44, 0xcc, 0x22, 0xaa, 0x66, 0xee, 0x11, 0x99, 0x55, 0xdd, 0x33, 0xbb, 0x77, 0xff };
uchar *dst = base;
const uchar *src = array;
int rowBytesSrc = (w+7)>>3 ;
int rowPatch = (pmp->rowBytes&0x3fff) - rowBytesSrc;
for ( int j=0; j<h; j++,dst+=rowPatch )
for ( int i=0; i<rowBytesSrc; i++,src++ )
*dst++ = (reverse[*src & 0x0f] & 0xf0) | (reverse[(*src >> 4) & 0x0f] & 0x0f);
}
}
UnlockPixels( pm );
}
}
SetPort(savePort);
return gw; /* tell caller we succeeded! */
}
void fl_delete_bitmask(Fl_Bitmask id) {
if (id) DisposeGWorld(id);
}
#elif defined(__APPLE_QUARTZ__)
Fl_Bitmask fl_create_bitmask(int w, int h, const uchar *array) {
static uchar reverse[16] = /* Bit reversal lookup table */
{ 0x00, 0x88, 0x44, 0xcc, 0x22, 0xaa, 0x66, 0xee,
0x11, 0x99, 0x55, 0xdd, 0x33, 0xbb, 0x77, 0xff };
int rowBytes = (w+7)>>3 ;
uchar *bmask = (uchar*)malloc(rowBytes*h), *dst = bmask;
const uchar *src = array;
for ( int i=rowBytes*h; i>0; i--,src++ ) {
*dst++ = ((reverse[*src & 0x0f] & 0xf0) | (reverse[(*src >> 4) & 0x0f] & 0x0f))^0xff;
}
CGDataProviderRef srcp = CGDataProviderCreateWithData( 0L, bmask, rowBytes*h, 0L);
CGImageRef id = CGImageMaskCreate( w, h, 1, 1, rowBytes, srcp, 0L, false);
CGDataProviderRelease(srcp);
return (Fl_Bitmask)id;
}
void fl_delete_bitmask(Fl_Bitmask id) {
if (id) CGImageRelease((CGImageRef)id);
}
#elif defined(WIN32) // Windows bitmask functions...
// 'fl_create_bitmap()' - Create a 1-bit bitmap for drawing...
static Fl_Bitmask fl_create_bitmap(int w, int h, const uchar *data) {
// we need to pad the lines out to words & swap the bits
// in each byte.
int w1 = (w+7)/8;
int w2 = ((w+15)/16)*2;
uchar* newarray = new uchar[w2*h];
const uchar* src = data;
uchar* dest = newarray;
Fl_Bitmask id;
static uchar reverse[16] = /* Bit reversal lookup table */
{ 0x00, 0x88, 0x44, 0xcc, 0x22, 0xaa, 0x66, 0xee,
0x11, 0x99, 0x55, 0xdd, 0x33, 0xbb, 0x77, 0xff };
for (int y=0; y < h; y++) {
for (int n = 0; n < w1; n++, src++)
*dest++ = (uchar)((reverse[*src & 0x0f] & 0xf0) |
(reverse[(*src >> 4) & 0x0f] & 0x0f));
dest += w2-w1;
}
id = CreateBitmap(w, h, 1, 1, newarray);
delete[] newarray;
return id;
}
// 'fl_create_bitmask()' - Create an N-bit bitmap for masking...
Fl_Bitmask fl_create_bitmask(int w, int h, const uchar *data) {
// this won't work when the user changes display mode during run or
// has two screens with differnet depths
Fl_Bitmask id;
static uchar hiNibble[16] =
{ 0x00, 0x80, 0x40, 0xc0, 0x20, 0xa0, 0x60, 0xe0,
0x10, 0x90, 0x50, 0xd0, 0x30, 0xb0, 0x70, 0xf0 };
static uchar loNibble[16] =
{ 0x00, 0x08, 0x04, 0x0c, 0x02, 0x0a, 0x06, 0x0e,
0x01, 0x09, 0x05, 0x0d, 0x03, 0x0b, 0x07, 0x0f };
int np = GetDeviceCaps(fl_gc, PLANES); //: was always one on sample machines
int bpp = GetDeviceCaps(fl_gc, BITSPIXEL);//: 1,4,8,16,24,32 and more odd stuff?
int Bpr = (bpp*w+7)/8; //: bytes per row
int pad = Bpr&1, w1 = (w+7)/8, shr = ((w-1)&7)+1;
if (bpp==4) shr = (shr+1)/2;
uchar *newarray = new uchar[(Bpr+pad)*h];
uchar *dst = newarray;
const uchar *src = data;
for (int i=0; i<h; i++) {
// This is slooow, but we do it only once per pixmap
for (int j=w1; j>0; j--) {
uchar b = *src++;
if (bpp==1) {
*dst++ = (uchar)( hiNibble[b&15] ) | ( loNibble[(b>>4)&15] );
} else if (bpp==4) {
for (int k=(j==1)?shr:4; k>0; k--) {
*dst++ = (uchar)("\377\360\017\000"[b&3]);
b = b >> 2;
}
} else {
for (int k=(j==1)?shr:8; k>0; k--) {
if (b&1) {
*dst++=0;
if (bpp>8) *dst++=0;
if (bpp>16) *dst++=0;
if (bpp>24) *dst++=0;
} else {
*dst++=0xff;
if (bpp>8) *dst++=0xff;
if (bpp>16) *dst++=0xff;
if (bpp>24) *dst++=0xff;
}
b = b >> 1;
}
}
}
dst += pad;
}
id = CreateBitmap(w, h, np, bpp, newarray);
delete[] newarray;
return id;
}
#if 0 // This doesn't appear to be used anywhere...
Fl_Bitmask fl_create_bitmask(int w, int h, const uchar *data, int for_mask) {
// we need to pad the lines out to words & swap the bits
// in each byte.
int w1 = (w+7)/8;
int w2 = ((w+15)/16)*2;
uchar* newarray = new uchar[w2*h];
const uchar* src = data;
uchar* dest = newarray;
Fl_Bitmask id;
static uchar reverse[16] = /* Bit reversal lookup table */
{ 0x00, 0x88, 0x44, 0xcc, 0x22, 0xaa, 0x66, 0xee,
0x11, 0x99, 0x55, 0xdd, 0x33, 0xbb, 0x77, 0xff };
for (int y=0; y < h; y++) {
for (int n = 0; n < w1; n++, src++)
*dest++ = (reverse[*src & 0x0f] & 0xf0) |
(reverse[(*src >> 4) & 0x0f] & 0x0f);
dest += w2-w1;
}
id = CreateBitmap(w, h, 1, 1, newarray);
delete[] newarray;
return (id);
}
# endif // 0
void fl_delete_bitmask(Fl_Bitmask bm) {
DeleteObject((HGDIOBJ)bm);
}
#else // X11 bitmask functions
Fl_Bitmask fl_create_bitmask(int w, int h, const uchar *data) {
return XCreateBitmapFromData(fl_display, fl_window, (const char *)data,
(w+7)&-8, h);
}
void fl_delete_bitmask(Fl_Bitmask bm) {
fl_delete_offscreen((Fl_Offscreen)bm);
}
#endif // __APPLE__
// MRS: Currently it appears that CopyDeepMask() does not work with an 8-bit alpha mask.
// If you want to test/fix this, uncomment the "#ifdef __APPLE__" and comment out
// the "#if 0" here. Also see Fl_Image.cxx for a similar check...
//#ifdef __APPLE_QD__
#if 0
// Create an 8-bit mask used for alpha blending
Fl_Bitmask fl_create_alphamask(int w, int h, int d, int ld, const uchar *array) {
Rect srcRect;
srcRect.left = 0; srcRect.right = w;
srcRect.top = 0; srcRect.bottom = h;
GrafPtr savePort;
GetPort(&savePort); // remember the current port
Fl_Bitmask gw;
NewGWorld( &gw, 8, &srcRect, 0L, 0L, 0 );
PixMapHandle pm = GetGWorldPixMap( gw );
if ( pm )
{
LockPixels( pm );
if ( *pm )
{
uchar *base = (uchar*)GetPixBaseAddr( pm );
if ( base )
{
PixMapPtr pmp = *pm;
// verify the parameters for direct memory write
if ( pmp->pixelType == 0 || pmp->pixelSize == 8 || pmp->cmpCount == 1 || pmp->cmpSize == 8 )
{
// Copy alpha values from the source array to the pixmap...
array += d - 1;
int rowoffset = (pmp->rowBytes & 0x3fff) - w;
for (int y = h; y > 0; y --, array += ld, base += rowoffset) {
for (int x = w; x > 0; x --, array += d) {
*base++ = 255 /*255 - *array*/;
}
}
}
}
UnlockPixels( pm );
}
}
SetPort(savePort);
return gw; /* tell caller we succeeded! */
}
#else
// Create a 1-bit mask used for alpha blending
Fl_Bitmask fl_create_alphamask(int w, int h, int d, int ld, const uchar *array) {
Fl_Bitmask mask;
int bmw = (w + 7) / 8;
uchar *bitmap = new uchar[bmw * h];
uchar *bitptr, bit;
const uchar *dataptr;
int x, y;
static uchar dither[16][16] = { // Simple 16x16 Floyd dither
{ 0, 128, 32, 160, 8, 136, 40, 168,
2, 130, 34, 162, 10, 138, 42, 170 },
{ 192, 64, 224, 96, 200, 72, 232, 104,
194, 66, 226, 98, 202, 74, 234, 106 },
{ 48, 176, 16, 144, 56, 184, 24, 152,
50, 178, 18, 146, 58, 186, 26, 154 },
{ 240, 112, 208, 80, 248, 120, 216, 88,
242, 114, 210, 82, 250, 122, 218, 90 },
{ 12, 140, 44, 172, 4, 132, 36, 164,
14, 142, 46, 174, 6, 134, 38, 166 },
{ 204, 76, 236, 108, 196, 68, 228, 100,
206, 78, 238, 110, 198, 70, 230, 102 },
{ 60, 188, 28, 156, 52, 180, 20, 148,
62, 190, 30, 158, 54, 182, 22, 150 },
{ 252, 124, 220, 92, 244, 116, 212, 84,
254, 126, 222, 94, 246, 118, 214, 86 },
{ 3, 131, 35, 163, 11, 139, 43, 171,
1, 129, 33, 161, 9, 137, 41, 169 },
{ 195, 67, 227, 99, 203, 75, 235, 107,
193, 65, 225, 97, 201, 73, 233, 105 },
{ 51, 179, 19, 147, 59, 187, 27, 155,
49, 177, 17, 145, 57, 185, 25, 153 },
{ 243, 115, 211, 83, 251, 123, 219, 91,
241, 113, 209, 81, 249, 121, 217, 89 },
{ 15, 143, 47, 175, 7, 135, 39, 167,
13, 141, 45, 173, 5, 133, 37, 165 },
{ 207, 79, 239, 111, 199, 71, 231, 103,
205, 77, 237, 109, 197, 69, 229, 101 },
{ 63, 191, 31, 159, 55, 183, 23, 151,
61, 189, 29, 157, 53, 181, 21, 149 },
{ 254, 127, 223, 95, 247, 119, 215, 87,
253, 125, 221, 93, 245, 117, 213, 85 }
};
// Generate a 1-bit "screen door" alpha mask; not always pretty, but
// definitely fast... In the future we may be able to support things
// like the RENDER extension in XFree86, when available, to provide
// true RGBA-blended rendering. See:
//
// http://www.xfree86.org/~keithp/render/protocol.html
//
// for more info on XRender...
//
// MacOS already provides alpha blending support and has its own
// fl_create_alphamask() function...
memset(bitmap, 0, bmw * h);
for (dataptr = array + d - 1, y = 0; y < h; y ++, dataptr += ld)
for (bitptr = bitmap + y * bmw, bit = 1, x = 0; x < w; x ++, dataptr += d) {
if (*dataptr > dither[x & 15][y & 15])
*bitptr |= bit;
if (bit < 128) bit <<= 1;
else {
bit = 1;
bitptr ++;
}
}
mask = fl_create_bitmask(w, h, bitmap);
delete[] bitmap;
return (mask);
}
#endif // __APPLE__
void Fl_Bitmap::draw(int XP, int YP, int WP, int HP, int cx, int cy) {
if (!array) {
draw_empty(XP, YP);
return;
}
// account for current clip region (faster on Irix):
int X,Y,W,H; fl_clip_box(XP,YP,WP,HP,X,Y,W,H);
cx += X-XP; cy += Y-YP;
// clip the box down to the size of image, quit if empty:
if (cx < 0) {W += cx; X -= cx; cx = 0;}
if ((cx+W) > w()) W = w()-cx;
if (W <= 0) return;
if (cy < 0) {H += cy; Y -= cy; cy = 0;}
if ((cy+H) > h()) H = h()-cy;
if (H <= 0) return;
#ifdef WIN32
if (!id) id = fl_create_bitmap(w(), h(), array);
HDC tempdc = CreateCompatibleDC(fl_gc);
int save = SaveDC(tempdc);
SelectObject(tempdc, (HGDIOBJ)id);
SelectObject(fl_gc, fl_brush());
// secret bitblt code found in old MSWindows reference manual:
BitBlt(fl_gc, X, Y, W, H, tempdc, cx, cy, 0xE20746L);
RestoreDC(tempdc, save);
DeleteDC(tempdc);
#elif defined(__APPLE_QD__)
if (!id) id = fl_create_bitmask(w(), h(), array);
GrafPtr dstPort;
GetPort( &dstPort );
Rect src, dst;
GetPortBounds( (Fl_Offscreen)id, &src );
SetRect( &src, cx, cy, cx+W, cy+H );
SetRect( &dst, X, Y, X+W, Y+H );
CopyBits(GetPortBitMapForCopyBits((Fl_Offscreen)id), // srcBits
GetPortBitMapForCopyBits(dstPort), // dstBits
&src, // src bounds
&dst, // dst bounds
srcOr, // mode
0L); // mask region
#elif defined(__APPLE_QUARTZ__)
if (!id) id = fl_create_bitmask(w(), h(), array);
if (id && fl_gc) {
CGRect rect = { { X, Y }, { W, H } };
Fl_X::q_begin_image(rect, cx, cy, w(), h());
CGContextDrawImage(fl_gc, rect, (CGImageRef)id);
Fl_X::q_end_image();
}
#else
if (!id) id = fl_create_bitmask(w(), h(), array);
XSetStipple(fl_display, fl_gc, id);
int ox = X-cx; if (ox < 0) ox += w();
int oy = Y-cy; if (oy < 0) oy += h();
XSetTSOrigin(fl_display, fl_gc, ox, oy);
XSetFillStyle(fl_display, fl_gc, FillStippled);
XFillRectangle(fl_display, fl_window, fl_gc, X, Y, W, H);
XSetFillStyle(fl_display, fl_gc, FillSolid);
#endif
}
/**
The destructor free all memory and server resources that are used by
the bitmap.
*/
Fl_Bitmap::~Fl_Bitmap() {
uncache();
if (alloc_array) delete[] (uchar *)array;
}
void Fl_Bitmap::uncache() {
if (id) {
fl_delete_bitmask((Fl_Offscreen)id);
id = 0;
}
}
void Fl_Bitmap::label(Fl_Widget* widget) {
widget->image(this);
}
void Fl_Bitmap::label(Fl_Menu_Item* m) {
Fl::set_labeltype(_FL_IMAGE_LABEL, labeltype, measure);
m->label(_FL_IMAGE_LABEL, (const char*)this);
}
Fl_Image *Fl_Bitmap::copy(int W, int H) {
Fl_Bitmap *new_image; // New RGB image
uchar *new_array; // New array for image data
// Optimize the simple copy where the width and height are the same...
if (W == w() && H == h()) {
new_array = new uchar [H * ((W + 7) / 8)];
memcpy(new_array, array, H * ((W + 7) / 8));
new_image = new Fl_Bitmap(new_array, W, H);
new_image->alloc_array = 1;
return new_image;
}
if (W <= 0 || H <= 0) return 0;
// OK, need to resize the image data; allocate memory and
uchar *new_ptr, // Pointer into new array
new_bit, // Bit for new array
old_bit; // Bit for old array
const uchar *old_ptr; // Pointer into old array
int sx, sy, // Source coordinates
dx, dy, // Destination coordinates
xerr, yerr, // X & Y errors
xmod, ymod, // X & Y moduli
xstep, ystep; // X & Y step increments
// Figure out Bresenheim step/modulus values...
xmod = w() % W;
xstep = w() / W;
ymod = h() % H;
ystep = h() / H;
// Allocate memory for the new image...
new_array = new uchar [H * ((W + 7) / 8)];
new_image = new Fl_Bitmap(new_array, W, H);
new_image->alloc_array = 1;
memset(new_array, 0, H * ((W + 7) / 8));
// Scale the image using a nearest-neighbor algorithm...
for (dy = H, sy = 0, yerr = H, new_ptr = new_array; dy > 0; dy --) {
for (dx = W, xerr = W, old_ptr = array + sy * ((w() + 7) / 8), sx = 0, new_bit = 1;
dx > 0;
dx --) {
old_bit = (uchar)(1 << (sx & 7));
if (old_ptr[sx / 8] & old_bit) *new_ptr |= new_bit;
if (new_bit < 128) new_bit <<= 1;
else {
new_bit = 1;
new_ptr ++;
}
sx += xstep;
xerr -= xmod;
if (xerr <= 0) {
xerr += W;
sx ++;
}
}
if (new_bit > 1) new_ptr ++;
sy += ystep;
yerr -= ymod;
if (yerr <= 0) {
yerr += H;
sy ++;
}
}
return new_image;
}
//
// End of "$Id$".
//